Biostat 1000 and Daugirdas blood-based hemodialysis quantification: agreement and reproducibility

Agreement and reproducibility of Daugirdas blood-based and Biostat 1000 dialysate-based Kt/V estimation were explored. Fifty-two dialysis treatments in 19 patients were studied. All patients were dialyzed by arteriovenous (AV) access. Good agreement was found in the comparison between laboratory predialysis blood urea nitrogen (BUN) and Biostat 1000 BUN. Each treatment was assessed for Kt/V simultaneously by Biostat 1000 and by Daugirdas methods based on predialysis and postdialysis BUN. Four Daugirdas blood-based Kt/V estimations per session were obtained; two were single pool Kt/V, the first using an "arterial" postdialysis BUN and the second a "mixed venous" postdialysis BUN, whereas the other two were double pool (or equilibrated) eKt/V obtained by factoring the respective single pool "arterial" and "mixed venous" Kt/V for the relative rate of solute removal. The four blood-based and Biostat 1000 Kt/V were examined for pooled-within-patient variability in 15 of the patients in whom three dialysis sessions on the same dialysis prescription were available, and these were not significantly different between the blood-based and Biostat 1000 Kt/V. The four blood-based Kt/V were then compared with the Biostat 1000 Kt/V using the concordance correlation coefficient (CC, 1 indicating pairs of observations fall on a line of identity, 0 indicating no relationship), and bias and range of agreement as defined by the Bland and Altman technique of analysis. The "mixed venous" eKt/V had the closest agreement with the Biostat 1000 Kt/V (CC = 0.77), but the range of agreement as defined by Bland and Altman was 0.62, implying that for a single session, there was a 95% chance that the "mixed venous" eKt/V would lie within +/- 0.31 of the Biostat 1000 Kt/V. It is concluded that Biostat 1000 Kt/V results are comparable in large groups to certain Daugirdas blood-based Kt/V, although for a given dialysis session, clinically important differences in resulting Kt/V parameters may be seen between these two methods of estimating Kt/V.